Father to Russell, Former CEO of Agio (exited), Authored two books, Oklahoma native, OU supporter, Yale MBA, aspiring golfer, living on the coasts and mountains
Joined January 2008
- Tweets 9,460
- Following 901
- Followers 2,665
- Likes 27,392
1,423 Photos and videos
Jun 10
Power users are the real Codex benchmark.
Give them higher limits.
Give them messy work.
Watch where they hit the wall.
That tells you more than another clean demo.
30
This is going to be a game changer for so many regular iPhone users who don't password variety nor change their passwords when a password is compromised.
46
I tried enabling the new Microsoft Scout agent (@MicrosoftAI) - spent over a day, used several agents to assist and still not working - this is the problem with @Microsoft right now - the agents experience is far too difficult to make work; where @OpenAI agents - JUST WORK.
49
You can outsource thinking, but you cannot outsource understanding.
1
55
May 31
My bias on voice agents:
fewer words
more completed work
Start the task.
Approve the action.
Fix the mistake.
Move on.
The best version probably means touching five fewer screens. x.com/OpenAIDevs/status/2060…
May 30
🧵 Our Voice Hack Night finalists are here.
4 projects. 6 hours. Realtime voice agents in real-world builds.
Now it’s your turn to vote for your favorite. We’ll announce the winner on Monday.
cerebralvalley.ai/e/openai-v…
1
80
May 30
Codex release notes are fun because each update makes the loop more real.
Windows computer use.
Mobile steering.
Usage stats.
Not just “better code.”
More like: the work follows you, keeps running, and starts to feel like an operator surface. x.com/OpenAIDevs/status/2060…
May 29
Codex now supports more of the Windows developer loop.
With Computer use on Windows, Codex can test apps, debug flows, and review work where your project context lives. Codex in the ChatGPT mobile app lets you connect to Windows machines and keep steering from your phone.
1
116
May 28
Private MCP workload identity is boring in the best way.
This is how agents get approved inside companies:
internal tools stay internal
auth is sane
security doesn’t have to pretend every demo is production-ready x.com/OpenAIDevs/status/2059…
May 27
Private MCP servers 🤝 OpenAI products
Your team can keep MCP servers inside your network while ChatGPT, Codex, and the Responses API connect through outbound-only HTTPS.
🔗 developers.openai.com/api/do…
1
1
63
May 27
AI thought of the day - while you can outsource knowledge and even some "thinking" - you can't outsource understanding.
22
May 25
Agent security is not only prompt injection.
It is tools, permissions, files, browser access, package scripts, CI runners, credentials, memory, and the agent confidently doing the wrong thing at machine speed.
The blast radius moved. The security model has to move with it.
2
82
May 25
Companies should pick one workflow and make it 30% cheaper, faster, or more reliable with AI.
Not a strategy deck.
Not a lab.
One workflow with a before and after.
1
33
May 25
Codex /goal is making me rethink what a prompt is.
A prompt asks for an answer.
A goal hands off work.
That sounds small until you start asking for outcomes instead of steps.
1
49
May 25
Codex /goal is easiest to underrate if you think in tasks.
The better frame is outcome handoff:
review this system
find the drag
make it simpler
check the edge cases
show me what changed
That is a very different workflow. x.com/daniel_mac8/status/205…
May 24
This is amazing. Do this.
Use Codex /goal mode and ask:
"Do a comprehensive review of my project and list areas to improve, simplify and make it more faithful to the plan."
1
57
May 24
AI as a force multiplier is a much better framing.
Not “let’s go do AI.”
More like:
take the workflow you already run every day, remove the drag, compress the cycle time, improve the review step.
That is where the money is.
22
May 24
Most companies should stop asking “what is our AI strategy?”
Ask this instead:
What process is too slow, too expensive, or too dependent on one person?
Start there. Use AI to make that process cheaper, faster, better.
1
24
May 24
Goal mode changes how I use Codex.
I stop asking for little tasks.
I start handing it outcomes:
fix this workflow, build the first version, check the edge cases, come back with what changed.
That is a different product.
25
May 24
AI vuln discovery just outpaced patching.
That should change how security teams plan.
The next bottleneck is not finding the bug.
It is deciding which reports matter and getting patches shipped.
1
19
May 23
One thing becoming clear with coding agents:
The model is less often the bottleneck.
The loop is context in, goal clarity, tool permissions, review surface, and recovery when it goes sideways.
That is the work now.
1
1
31
May 22
Agent skills are going to need the same seriousness we eventually learned with packages and CI.
Source matters. Permissions matter. Updates matter.
“Install this skill” should not become the new “just run this script.” x.com/NVIDIAAI/status/205749…
We just shipped NVIDIA-Verified Agent Skills 🔐
Skills make your agent more capable, but can also introduce vulnerabilities. Verified skills give you transparency into what a skill does, where it came from, what risks it carries, and whether it's been modified.
Every verified skill carries a skill card and is built on the agentskills.io open specification to work reliably across @claudeai Code, @openai Codex, and @cursor_ai.
46
May 22
Codex Thursday is basically the workflow filling in.
Appshots. Goal mode. Locked Mac use. Plugins. Better analytics.
Less “chat with a model.”
More “hand work to an agent and check back.” x.com/OpenAI/status/20576178…
Highlights from today’s Codex Thursday launches:
1️⃣ Codex can now securely use apps on your Mac from your phone, even when your Mac is locked and the screen is off.
developers.openai.com/codex/…
1
112
May 20
This matches my usage.
Codex feels less like autocomplete and more like a patient repo engineer.
It reads the room better: surrounding files, tests, edge cases, risk.
Claude Code still has moments, but Codex is the one I trust with bigger changes right now.
65